|Publication number||US5379406 A|
|Application number||US 08/178,485|
|Publication date||Jan 3, 1995|
|Filing date||Jan 7, 1994|
|Priority date||Aug 5, 1988|
|Also published as||US5067107|
|Publication number||08178485, 178485, US 5379406 A, US 5379406A, US-A-5379406, US5379406 A, US5379406A|
|Inventors||Gerald T. Wade|
|Original Assignee||Hewlett-Packard Company|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (3), Referenced by (33), Classifications (8), Legal Events (6)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application is a continuation of co-pending application Ser. No. 07/793,740, filed Nov. 18, 1991, now abandoned, which was a continuation of co-pending application Ser. No. 07/229,573 filed Aug. 5, 1988, which issued as U.S. Pat. No. 5,067,107 on Nov. 19, 1991.
The present invention relates to the measurement of computer performance.
Performance tools are used to measure the performance of a computer. The performance of a computer may be evaluated for a variety of purposes. For instance, it may be desired to make a "quick scan" of a computer system. What is meant by a quick scan is an examination of what is currently happening on the computer system in "real time" A quick scan may be done, for instance, when system performance is poor and it is desired to identify which processes are utilizing which computer resources and thus causing the poor performance.
Alternately, a determination of the "state of the system" may be done in order to evaluate the total resource utilization of the system. Such a determination may be done to discover which resources cause slow downs or bottlenecks in system performance. Once identified these resources may be upgraded to improve system performance.
Another reason for evaluating computer performance may be for "application tuning", that is to focus on a particular user application or situation to determine how to improve system performance as regards that particular application or situation.
Finally, performance tools may be used for "trouble shooting", that is determining why system performance degrades for no immediately apparent reason.
In the prior art, performance tools have been written for sophisticated computer technicians to aid in trouble shooting. The general philosophy has been the more data that is generated the better. When these performance tools have been adapted for other uses such as for quick scans, determination of the state of the system or for application tuning, the performance tools have generated too much information. This information is often incomprehensible to the average computer user.
In some system attempts have been made to arrange and to present gathered data in some form which may be deciphered by a user. However, such systems have not proved practical for gathering and presenting data for system performance over a long period of time. This is because the programs typically log an immense amount of information, before the information is later reduced. The memory resources required for such data logging are inordinate. A typical prior art performance system may log one hundred twenty million bytes in a typical twenty-four hour period. Not only is this burdensome on memory resources, but it also takes a lot of processing power to generate this information and then to manipulate the this data into a form which is useful to a user.
In accordance with the preferred embodiments of the present invention a performance and measurement system for a computing system is presented. Performance data produced by the computing system is collected and reduced before being logged. Once the data is logged, the data may be transported to a workstation and accessed by a user.
In the preferred embodiment, the collected and reduced data is logged into three files. The first file is a global data file into which is logged information about the computing system as a whole. The second file is a process data file into which is logged information about selected processes. The third file is a workload data file into which is logged information about classes of processes.
Information about processes are logged into the process data file when a process does something "interesting" during a specified interval. For example, a process may be considered to have done something interesting when the process is started , is terminated or uses a predetermined amount of a computing system resource during the specified interval.
Data volume is reduced by combining available system performance numbers into standard metrics (See Appendix B for examples of standard metrics). By performing this reduction prior to logging the data, significant savings in log file space are achieved. Contrary to expectations, the pre-reduction of data before logging results in less overhead than that which is required to log the data without pre-reduction. Choice of standard performance metrics may be made so that they are understandable to a user and not specific to a particular computing system.
By selecting to log information only about "interesting" processes, the amount of information logged also is greatly reduced. Since many of a system's processes are dormant or almost dormant at any one time, logged information about these processes would be of little interest when studying the performance of a computing system.
FIG. 1 is a block diagram representation of typical prior system used for evaluating computer performance.
FIG. 2 is a block diagram of a system for computer performance measurement and ev-aluation in accordance with the preferred embodiment of the present invention.
In FIG. 1 a block diagram of a typical prior art computer performance system is shown. A computer system 11 produces performance data 12 which is logged in a logging file 13. When a user wishes to examine system performance, data 14 is transferred from logging file 13 to a filtering program 15. Filtering program 15 arranges the data in a prescribed format and forwards filtered data 16 to a user workstation 17 for evaluation by the user.
FIG. 2 shows a computer performance measurement and evaluation system in accordance with the preferred embodiment of the present invention. Computer system 11 is again shown producing performance data 12. This information is selectively collected and reduced to standard metrics by data collection and reduction program 21. For an example of standard metrics see Appendix B.
The most commonly used techniques for producing performance data are (1) Sampling, (2) Tracing and (3) Counting. When sampling takes place, computing system 11 is periodically interrupted and the current status of the system is determined. This provides snapshots of the system which may be statistically analyzed to evaluate the system performance.
Tracing is done by the insertion of code into the operating system of computing system 11 which causes the recording of key events. Typically each record of a key event will include a time stamp showing precisely when the event occurred. For example, if the initialization of a process and the termination of a process are recorded as key events, it is possible to determine how long the process was running.
Counting is done by the insertion of code into the operating system, much like tracing. The difference is that the inserted code increments a pre-defined accumulator or accumulates elapsed time into such an accumulator each time an event occurs, rather than recording the event itself.
Data collection and reduction program 21 receives the produced data 12 and will reduce the data before any logging takes place. Program 21 groups the reduced data into three groups and logs different details for each group. The first group is global data logged in global data file 23. The second group is process data logged in process data file 24. The third group is workload data logged in workload data file 25.
Global data in global data file 23 consists of global data records of standard metrics which report system performance on computing system 11 as a whole. Records are added to this file, for example, at the rate of one every five minutes.
Process data in process data file 24 consists of process data records of standard metrics which contains information on any process which is found to be "interesting" during a specified interval. Specified interval is, for example, one minute. Records are typically added to process data file 24 at the end of each specified interval. A process is found to be "interesting" when it meets any one (or more) of the following criteria:
A process is seen for the first time, i.e., it is started.
A process ceases to run, i.e., it is terminated.
A process uses more of a given resource than a user defined threshold value. For instance, thresholds may be set for (1) central processing unit (CPU) usage, (2) amount a disc drive is accessed, (3) amount of transactions with a terminal and(4) the amount of response time between the process and a terminal.
Workload data in workload data file 25 consists of records containing information for any user defined grouping or processes. Such a grouping is called a "class". For example, information about a class may be logged once every five minutes when at least one process within the class consumes some CPU time during the five minutes. An extensive level of detail may be logged about a class of processes where it would be impractical to log such a level of detail for each process.
In the preferred embodiment, a user may specify whether to log global, process and workload data independently or in any combination. A user also may set the amount of disc space the user is willing to consume to hold data of each type. Global data file 23, process data file 24 and workload data file 25 are "circular", that is when each file is filled new data overwrites the oldest data in the file.
Table 1 below presents a "pseudo code" version of data collection and reduction program 21 .
TABLE 1______________________________________INITIALIZE (Verify execution environment and initialize counters) (Open global data file 23, process data file 24, workload data file 25)READ USER DEFINABLE PARAMETERS (System Identifi- cation, Thresholds, Application Definitions)BEGIN (Continuous Loop) GATHER SYSTEM GLOBAL INFORMATION (From Operating System) COMBINE RAW GLOBAL DATA INTO STANDARD GLOBAL METRICS BEGIN (For each potential Process) GATHER PROCESS INFORMATION (From Operating System) ACCUMULATE PROCESS INFORMATION INTO STANDARD GLOBAL METRICS IF THIS IS A NEW PROCESS THEN MARK THE PROCESS AS "INTERESTING" DETERMINE APPLICATION # FROM USER DEFINITIONS IF THIS PROCESS JUST TERMINATED THEN MARK THE PROCESS AS "INTERESTING" IF THIS PROCESS EXCEEDED ANY USER DEFINED THRESHOLDS (CPU%, DISC IO RATE, TRANSACTION RATE, RESPONSE TIME) MARK THE PROCESS AS "INTERESTING ACCUMULATE PROCESS INFORMATION INTO CORRESPONDING APPLICATION'S STANDARD METRICS AND MARK THE APPLICATION AS "ACTIVE" IF THE PROCESS IS MARKED "INTERESTING" THEN COMBINE RAW PROCESS DATA INTO STANDARD PROCESS METRICS LOG THE PROCESS METRICS INTO process data file 24. END IF THE DESIRED GLOBAL INTERVAL HAS PASSED THEN LOG THE GLOBAL METRICS INTO global data file 23. FOR EACH USER DEFINED APPLICATION IF APPLICATION IS MARKED AS "ACTIVE" THEN LOG THE APPLICATION METRICS INTO workload data file 25. INITIALIZE APPLICATION COUNTERS PAUSE UNTIL TIME TO COLLECT NEXT DATAEND______________________________________
In order to present the collected data to a user, it is necessary to transport the data from files 23, 24 and 25 to user workstation 17. This may be done by a transport server 26. Transport server 26 may deliver the data to user workstation 17, for instance by a local area network (LAN) or an RS-232 data communication link. A data extract program 27 may be used to extract data from computers on a network which user workstation is not directly a part of.
Appendix A contains a description of a user interface for the computer performance and evaluation system of FIG. 2.
Appendix B contains an listing of standard performance metrics which may be logged into global data file 23, process data file 24 and workload data file 25. ##SPC1##
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4507751 *||Jun 21, 1982||Mar 26, 1985||International Business Machines Corporation||Method and apparatus for logging journal data using a log write ahead data set|
|US4878167 *||Jun 30, 1986||Oct 31, 1989||International Business Machines Corporation||Method for managing reuse of hard log space by mapping log data during state changes and discarding the log data|
|US5062055 *||May 29, 1990||Oct 29, 1991||Digital Equipment Corporation||Data processor performance advisor|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US5699505 *||Aug 20, 1996||Dec 16, 1997||Unisys Corporation||Method and system for automatically collecting diagnostic information from a computer system|
|US5764912 *||May 20, 1997||Jun 9, 1998||Advanced System Technologies||Method and apparatus for determining response time in computer applications|
|US5781449 *||Aug 10, 1995||Jul 14, 1998||Advanced System Technologies, Inc.||Response time measurement apparatus and method|
|US5838920 *||Jul 18, 1997||Nov 17, 1998||Advanced System Technologies, Inc.||Method and apparatus for identifying transactions|
|US5852713 *||Sep 30, 1997||Dec 22, 1998||Shannon; John P.||Computer data file backup system|
|US5887167 *||Nov 3, 1995||Mar 23, 1999||Apple Computer, Inc.||Synchronization mechanism for providing multiple readers and writers access to performance information of an extensible computer system|
|US5920719 *||Nov 6, 1995||Jul 6, 1999||Apple Computer, Inc.||Extensible performance statistics and tracing registration architecture|
|US6044335 *||Dec 23, 1997||Mar 28, 2000||At&T Corp.||Productivity metrics for application software systems|
|US6308211||Dec 16, 1998||Oct 23, 2001||Computer Associates Think, Inc.||Method and apparatus for identifying informational packets|
|US6470388||Jun 10, 1999||Oct 22, 2002||Cisco Technology, Inc.||Coordinated extendable system for logging information from distributed applications|
|US6477571||Aug 11, 1998||Nov 5, 2002||Computer Associates Think, Inc.||Transaction recognition and prediction using regular expressions|
|US6493754||Nov 5, 1998||Dec 10, 2002||Computer Associates Think, Inc.||Method and apparatus for analyzing communications on different threads|
|US6522631||Apr 9, 2002||Feb 18, 2003||Computer Associates Think, Inc.||Response time measurement apparatus and method|
|US6832342 *||Mar 1, 2001||Dec 14, 2004||International Business Machines Corporation||Method and apparatus for reducing hardware scan dump data|
|US6901536 *||Dec 3, 2001||May 31, 2005||Microsoft Corporation||Service quality monitoring system and method|
|US6990521||Oct 4, 2002||Jan 24, 2006||Computer Associates Think, Inc.||Transaction recognition and prediction using regular expressions|
|US6996807||Nov 27, 2000||Feb 7, 2006||Isogon Corporation||Consolidation and reduction of usage data|
|US7086057 *||May 15, 2002||Aug 1, 2006||Matsushita Electric Industrial Co., Ltd.||Task allocation time decision apparatus and method of deciding task allocation time|
|US7200651||Jul 2, 1999||Apr 3, 2007||Cisco Technology, Inc.||Dynamic configuration and up-dating of integrated distributed applications|
|US7483976||Nov 11, 2005||Jan 27, 2009||Computer Associates Think, Inc.||Transaction recognition and prediction using regular expressions|
|US7493527 *||May 24, 2005||Feb 17, 2009||International Business Machines Corporation||Method for logging diagnostic information|
|US7890473 *||Oct 26, 2004||Feb 15, 2011||Sprint Communications Company L.P.||Method and system for evaluating performance of a computer system|
|US8255511||Jul 28, 1999||Aug 28, 2012||Ca, Inc.||Network performance monitoring|
|US8326965||May 3, 2001||Dec 4, 2012||Hewlett-Packard Development Company, L.P.||Method and apparatus to extract the health of a service from a host machine|
|US20020124210 *||Mar 1, 2001||Sep 5, 2002||International Business Machines Corporation||Method and apparatus for reducing hardware scan dump data|
|US20020165892 *||May 3, 2001||Nov 7, 2002||Doug Grumann||Method and apparatus to extract the health of a service from a host machine|
|US20020174164 *||May 15, 2002||Nov 21, 2002||Kunihiko Hayashi||Task allocation time decision apparatus and method of deciding task allocation time|
|US20020198976 *||Dec 3, 2001||Dec 26, 2002||Microsoft Corporation||Service quality monitoring system and method|
|US20050278706 *||May 24, 2005||Dec 15, 2005||International Business Machines Corporation||System, method, and computer program product for logging diagnostic information|
|US20060031458 *||Oct 4, 2002||Feb 9, 2006||Platinum Technology, Inc||Transaction recognition and prediction using regular expressions|
|US20060064487 *||Nov 11, 2005||Mar 23, 2006||Computer Associates Think, Inc., A Delaware Corporation||Transaction recognition and prediction using regular exp|
|WO2001057670A2 *||Jan 26, 2001||Aug 9, 2001||Isogon Corporation||Consolidation and reduction of computer usage data|
|WO2001057670A3 *||Jan 26, 2001||Jan 3, 2002||Isogon Corp||Consolidation and reduction of computer usage data|
|U.S. Classification||703/21, 714/E11.204, 714/E11.192|
|Cooperative Classification||G06F11/3409, G06F11/3476|
|European Classification||G06F11/34T4, G06F11/34C|
|Jul 2, 1998||FPAY||Fee payment|
Year of fee payment: 4
|Jan 16, 2001||AS||Assignment|
Owner name: HEWLETT-PACKARD COMPANY, COLORADO
Free format text: MERGER;ASSIGNOR:HEWLETT-PACKARD COMPANY;REEL/FRAME:011523/0469
Effective date: 19980520
|Jul 2, 2002||FPAY||Fee payment|
Year of fee payment: 8
|Jul 23, 2002||REMI||Maintenance fee reminder mailed|
|Jul 3, 2006||FPAY||Fee payment|
Year of fee payment: 12
|Sep 22, 2011||AS||Assignment|
Owner name: HEWLETT-PACKARD DEVELOPMENT COMPANY, L.P., TEXAS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:HEWLETT-PACKARD COMPANY;REEL/FRAME:026945/0699
Effective date: 20030131